advertisement

Editors Selection IGR 14-1

Comments

Comment by Kouros Nouri-Mahdavi on:

24877 Three-year follow-up of the tube versus trabeculectomy study, Gedde SJ; Schiffman JC; Feuer WJ et al., American Journal of Ophthalmology, 2009; 148: 670-684

Find related abstracts

First, it is appropriate for me to acknowledge that I have a conflict of interest in considering the 'tube' versus 'trab' study that is the basis of this commentary, specifically, the 'Three-Year Follow-up of the Tube Versus Trabeculectomy Study' by Steven Gedde, Joyce Schiffman, et al. I served as a member of the Data and Safety Monitoring Committee. It is my belief that this study was well-designed, meticulously implemented and appropriately reported. In many ways, then, it stands as a model of what is generally considered the highest level of evidence-based medicine, a prospective, randomizedcontrolled clinical trial.

The 'tube versus trab' study reminds me of a similar randomizedcontrolled trial, of equally high quality, that was destined to have virtually no effect on clinical care; by and large, its findings have been meticulously ignored. I refer to the Glaucoma Laser Trial (GLT). The GLT appeared to demonstrate what almost no 'responsible' ophthalmologist wanted to hear, specifically that argon laser trabeculoplasty (ALT) controlled intraocular pressure as well as one method of employing topical anti-glaucoma medications. However, it was accompanied by an unfavorable editorial by a well-respected ophthalmologist who misread the report, it appeared to justify what an appropriately conservative population of physicians considered an excessively aggressive surgical adventure, and, in fairness to its critics, effective medications that were not available at the initiation of the study came on to the market, presumably changing the balance. One of the findings that was ignored was the decrease in the size of the cup/disc ratio, and the decrease in the mean defect of the visual field that was associated with treatment.

The generic problem that is an integral part of a study such as the GLT, and even more of the 'tube versus trab' study, is whether any study can answer a question, 'Which is better?' when the basis for that answer will be counting up subjective determinations which will always be different for the items being compared. For example, the GLT showed that argon laser trabeculoplasty lowered pressure as effectively as drops. But, there is an understandable perception that lasers are dangerous, and that drops are safe. After all, lasers can be used to blow missiles out of the sky! How does one compare the relative importance of the inconvenience and side effects of using drops with the anxiety of having your eye blown apart with a laser? Whenever trying to answer the question, 'Which is better?' the answer will be dissatisfying when the ways to measure which is better differ for the two things being considered. For example, when trabeculectomies are performed so that they cause immediate filtration, and tubes are performed so that the tube is tied off so that there is usually no immediate filtration, clearly there will be a higher incidence of flat anterior chambers in the trab group than in the tube group. On the other hand, double vision is rarely a complication of trabeculectomy, but, understandably, is a significant complication with tube shunt procedures. But how do you compare the significance of these complications with each other? For example, are ten flat anterior chambers about the same significance as one double vision? And does it take about ten double visions to have the same significance as one endophthalmitis? How does one meaningfully place objective values on these necessarily subjective considerations?

Additionally, the comparative effect of two interventions can be accurately determined only when the two interventions are the only two variables, and all the other considerations in the two groups are the same. It is true that one can use multi-variant analysis to shake out some associations, but the validity of such conclusions is suspect. For example, consider something as simple as trying to establish whether tubes or trabs lower intraocular pressure more effectively. The only way that can be answered is if all of the factors which might affect intraocular pressure are the same in the two groups. But this was not the case in the tube versus trab study. The investigators were permitted to use intraocular pressure-lowering medications post-operatively as they saw fit. One might say that it was not surprising that the pressures were roughly comparable, because the investigators used more medications in one group than the other group, so that the results would be comparable. The study was designed in such a fashion, because the subjects in the study were human beings, and the primary responsibility of the investigators was to make sure that those human beings got treatments that would give them the best chance of having a reasonably satisfactory outcome.

Despite criticisms about the dangers of 'data mining', it may be that the major value of many well-designed, well-executed controlled clinical trials comes from such explorations. For example, one might say that the most important value of the GLT was not that it showed that ALT worked as well or better than drops, but rather that pressure lowering could result in a decrease in cup/disc ratios and in the mean defect of visual fields. Furthermore, the Advanced Glaucoma Intervention Study did not settle the issue as to whether patients should be treated with ALT or trabeculectomy, but it showed that patients with consistently low intraocular pressures do better than patients with consistently higher intraocular pressures.It may be that the greatest value of the tube versus trab study will not be known until about ten years from now, and then, not because it will have answered whether tubes are better than trabs, but rather it will have established the long-term outcomes of each of these two procedures. It is likely that there will be significant differences in those outcomes. Those differences do not necessarily make one procedure better or worse, but they can give meaningful guidance to thoughtful physicians trying to select what is the most appropriate procedure for each individual patient.

• ← Previous
• Next →